Allard, M., Kasper, J.N.
Transcription
Allard, M., Kasper, J.N.
Th e Int 7th ern Per atio ma Co fro nal nfe st ren ce Yel Jun lowkni 199 e 23- fe 27 8 PERMAFROST - Seventh International Conference (Proceedings), Yellowknife (Canada), Collection Nordicana No 55, 1998 ? TEMPERATURE CONDITIONS FOR ICE-WEDGE CRACKING: FIELD MEASUREMENTS FROM SALLUIT, NORTHERN QUƒBEC Michel Allard, Jennifer N. Kasper Centre dÕŽtudes nordiques UniversitŽ Laval, Sainte-Foy, QuŽbec, Canada G1K 7P4 e-mail: michel.allard@cen.ulaval.ca Abstract The temperatures at which thermal cracking occurred along ice-wedges around a tundra polygon were measured over two years near Salluit, northern QuŽbec. Electrical cables were buried in the active layer across furrows or cracks in the soil at various places around the polygon. The time of breaking of the electrical cables, and the air, soil surface and ground temperatures down to 2.5 m were monitored with a datalogger. In the course of the two winters, several cables broke with the opening of thermal cracks. Over the two years, the first cracks opened in late December-early January when the temperature at the permafrost table was about - 15¡C, and after a drop of air temperature from about -20¡C to below -32¡C. Mean cracking temperature at wedge top was -20¡C in the first year and -19.7¡C in the second year. The data also allow estimates of the minimum temperature changes and cooling rates required to induce ice-wedge cracking. The cracks closed (or narrowed) and re-opened (or widened) in response to winter temperature fluctuations at the soil surface. The measured thermal conditions for cracking substantiate the previous theoretical work on this basic process at the origin of tundra polygons. surements of air and ground temperatures at the time and site of frost cracking have not yet been reported. Introduction Cracking of the soil, the process at the origin of icewedges and sand-wedges and of tundra polygons (Leffingwell, 1919), is a temperature-controlled phenomenon. In theoretical work on the mechanics of thermal contraction cracking, Lachenbruch (1962) estimated that the required value of soil viscosity to allow frost cracking is reached when air temperatures fall between -20 and -30¡C and when the temperature at the top of permafrost is about -15 to -20¡C (Lachenbruch, 1966; PŽwŽ, 1966). However, for cracking to take place, a rapid drop of temperature is the triggering mechanism. With a too low cooling rate, the tensile stress induced by thermal contraction can be released by permafrost creep. ÒLess than half a day of cooling... is required for a close approach to the full stress value when cooling rate is 10¡C/day (a total cooling of about 4¡C over a period of about 9 hours). Lower cooling rates must be maintained longer to be effectiveÓ (Lachenbruch, 1962, p. 20). Similarly, Grechishchev (1973, p. 231) concludes that: Òthe dynamics of the formation of frost fissures obviously look like the following: long period cooling leads to stress equal to the long term strength, and rupture is caused by secondary short-period temperature fluctuationsÓ. Detection of several hundreds of frost cracking events and careful examination of field and climatic conditions by Mackay (1974, 1978, 1984, 1992, 1993) supports these theoretical studies. However, mea- Over two winters, we measured the air and soil temperatures at which frost cracking occurred along a few ice-wedges around a polygon. This was done at a site that lies near the southern margin of the distribution range for active ice-wedges. Study area and characteristics of the experimental site The test site is in the Foucault river valley (Narsajuaq), 12.5 km west of Salluit (Figure 1). The mean annual air temperature for the area is estimated to be about -8¡C (Gray, 1983). At the study site, our measurements in 1990 indicated a mean air temperature of - 8.9¡C. The annual temperature range is about 37¡C. Total precipitation is estimated to be 310 mm, of which 53% is snow. Ground temperature at the depth of zero annual amplitude (23 m) is about -6.2¡C, as measured in drill holes near Salluit airport (Allard et al., 1995). The experimental site itself is on a fluvial terrace at an elevation of 14.5 m a.s.l. It bears a network of random orthogonal, low centre polygons with a mean diameter of 16 m (Figure 2). The soil consists of a 2-2.6 m thick sequence of organic-rich fine sand layers alternating with 10-30 cm thick sandy organic layers that overlie coarse fluvial and gravely sands. These fluvial sediments are themselves about Michel Allard, Jennifer N. Kasper 5 Methods INSTRUMENTATION Two thermistor cables (thermistors YSI-44033) were installed in drill-holes, one in the centre of the polygon, the other under a trough along a side, through an icewedge. The centre of the polygon is about 40 cm deeper than its raised edges and the trough along the polygon side is about 30 cm deep ; therefore the soil surface over the wedge is about 10 cm above the level of the polygon centre. The cables were set in PVC casings filled with silicone oil (Osterkamp, 1974). A thermistor was also installed on top of a 3 m high mast, in a shelter, to measure air temperatures, and one was at the ground surface (+/- 5 cm deep under the moss in polygon centre). Temperature readings were recorded by a datalogger (Campbell Scientific CR-10, 12 channels). Two channels were used to store hourly temperatures in the air and at the ground surface. Every four hours, temperatures were recorded at 55, 105 and 255 cm depths in the polygon centre and at 8, 58, 108, 158 and 258 cm depths in the polygon trough. Figure 1. Location of the experimental site. 7 m thick and they overlie post-glacial marine sandy silts. In river bluff sections and on ground probing radar profiles, the ice-wedges are seen to penetrate downward 5-6 m into the fluvial sediments (see Allard et al., 1993, p.7). According to the regional uplift curve, the site must have emerged about 4700 years ago, setting a potential date for the onset of frost cracking and initiation of the polygons (Kasper, 1995). As wedge ice was met in the drill hole in the trough at a depth of 33 cm, which is close to mean wedge depth in the region (62 measurements), a depth of 30 cm is used below as the wedge top depth for interpolations of temperature of cracking. The last available channel on the data logger was used for a cracking detection system which consisted of a main electric circuit including ten resistors of different values in series. Each resistor was bypassed with an Figure 2. Plan of the instrument layout. 6 The 7th International Permafrost Conference electrical cable loop. Incremental increase in resistance due to the breaking of one loop spread across a potential crack could be monitored with the datalogger. The resistance increment could be used to identify the broken loop. As the values of the resistors differed by equal increments and as the circuit was sensed only every four hours (a design fault found afterwards), there existed the potential situation that if two cables broke between two readings, the total resistance shift on the circuit equaled the one of a third different cable, leading to an identification mistake. For the first winter the structure of the data indicates that this did not occur. The loops, or breaking cables, were first laid across cracks on 12 July, 1989 at depths of about 15 cm in the thawing active layer. Nine of them extended across the top of known ice wedges (Figure 2). One was laid on the inner side of a small polygon ridge where, apparently, new cracking was taking place. The cables were dug up one year later on 1 July, 1990, with a small amount of damage resulting from excavating through 5-10 cm of frozen ground. The broken cables were soldered and buried again, taking care that an unaffected segment of cable extended across the wedge. For the second winter, jumps in resistance were clear-cut in the data, but identification of which cables broke on particular events was not always possible. DATA ANALYSIS During excavation in 1990, 7 of the 10 cables showed evidence of strain. As the type of electrical wire used in the loops was rather coarse (steel wires coated in translucent rubber sheeting of 2 mm o.d.), it could be seen that tension on some of them had broken the steel wire but that the rubber coating had been only stretched. As crack closing had taken place, some of them were still pinched in ice veins and in the cracks. On 21 August, 1991, re-excavation revealed that 8 cables had been affected and that their mechanical behavior had been the same as in the previous winter. Despite the coarse time interval for cracking detection (4 hours), discrete cracking events could be discerned. An unexpected outcome was that open cracks partially closed and re-widened several times during the winter, each time re-closing sub-circuits and registering shifts in the resistance channel, thus permitting assessment of both opening and closing events. The logged temperature data were quality checked for consistency and loaded into a spreadsheet software. Then, times and temperatures in the soil and in the air were determined for each cracking and each closing event. The peak or trough of the temperature oscillation preceding the event was searched both in the data table and on ÒT vs timeÓ graphs (Figures 3 and 4) to determine the range of the triggering temperature oscilla- tion, its duration and the rate of change (Table 1). A minus sign applied to a rate means, in the tables and in the text, a cooling rate. Results WINTER 1989-90 The steel wires in five cables broke between 27 December and 15 January (#7, #3, #1, #4 and #6 in sequence). The first rupture took place when the air temperature was -32.3¡C; ground temperatures were -17.5¡C at 8 cm depth, and -15.3¡C at the wedge top (interpolated). The vertical temperature gradient was 10¡C/m from the surface through the top of the wedge and diminished to 1.9¡C/m in the lower meter of the profile. This first cracking occurred during the first major cold spell of the winter which had begun 34 hours beforehand and after a cooling of 10.6¡C, that is at a rate of -0.3¡C/hr (-7.4¡C/day). At 8 cm, the cooling went back 140 hours at a rate ten times smaller and it went to 240 hours at 58 cm, at a very low cooling rate of -0.0096¡C/hr (-0.23¡C/day) (Table 1). This first cracking, and the second one which occurred in the same day, may have been a response to tensile stress at the soil surface given the small cooling rates at depths greater than 8 cm which are part of the general early winter cooling. All crack openings of this winter (including reopenings after January 15) occurred at air temperatures of -26.8¡C or colder (average of -31.5¡C), after a mean drop of 11.4¡C over a mean time of 34 hours (average rate of -0.6¡C/hr or -14.3¡C/day). A few cracks opened after a cooling time of only 4-10 hours. Openings took place when the temperature at 8 cm depth was below -17.5¡C (average of -21.7¡C) and after an average cooling period of 60 hours for a cooling rate of -0.07¡C/hr (-1.73¡C/day). The data also show that openings sometimes took place as the ground at 58 cm was warming up a little. This implies that tensile stresses induced at the soil surface and at the wedge top were provoked during falling air temperatures, as waves from warmer spells of a few days before were still propagating in the ground. We can interpolate that the average temperature for cracking at the wedge top was about -20.0¡C. At this level, openings occurred after a cooling of roughly 2.14¡C over an average time of 90 hours for an estimated rate of - 0.024¡C/hr (-0.57 ¡C/day). All closings took place during warming spells, with the exception of cable #4 on 20 January. This cable had opened 4 to 8 hours before and contact was somehow re-established. The average warming associated with closings is 11.5¡C over an average period of 30 hours. At 8 cm, the warming and the period were, respectively, 2.2¡C over 29.8 hours. But at 58 cm the average change was -0.205 ¡C over 65 hours (Table 1). The near surface nature of the mechanism is here again evident. Michel Allard, Jennifer N. Kasper 7 Figure 3. Air and soil temperature curves (1989-90 and 1990-91). Upward arrowheads : opening events. Downward arrowheads : closing events. The general picture for winter 1989-90 is rather simple. From the last week of December to the first week of March, air and near surface temperatures were constantly cold (mean air temperature of -28.2 ¡C for January , -30.7¡C for February, and -21.5¡C for March) with 4-5 days oscillations. The remainder of March had two warm spells and two cold ones. Openings and closings followed these oscillations within a rheologicaly suitable temperature range (Figure 3). 8 WINTER 1990-91 Wedge cracking started on January 2 when the air temperature was -34.5¡C, after a drop of 14.8¡C over 108 hours. Along the thermal profile in the wedge cable, the temperature was -16.6¡C at 8 cm and -14.1¡C at permafrost or wedge top (Table 1). It is worth noting that three openings took place when the air was warming, albeit at very low temperatures (average -27.2¡C). The crackings and the subsequent opening events all occurred below -24.4¡C for an average air temperature of -32.8 ¡C, and after a cooling of 13.1¡C The 7th International Permafrost Conference Table 1. Average and threshold thermal conditions for opening and closing of cracks See Conference CD-Rom for complete data files over 102.2 hours (a rate of -0.15¡C/hr or -3.7¡C/day). Just below the soil surface, the general conditions for cracking were a temperature of -21.6¡C, after a cooling of 4.6¡C over 134.7 hours for a rate -0.038¡C/hr (-0.9¡C/day). At the wedge top, the average temperature for crack openings is interpolated at - 19.7¡C ; the triggering drop of temperature was in average of 3.8¡C and lasted roughly 143 hours (a rate of -0.64 ¡C/day). Table 1 shows that the three closing events occurred at very cold air (-27¡C to -33¡C) and soil temperatures (-18.8¡C to - 25¡C). However they took place after much shorter temperature changes (27-48 hours) than did the openings. The temperature curves for this winter depict a very different history from the previous year (Figure 4). January was colder by 3.7¡C (mean of - 31.9¡C), February was warmer by 3.5¡C (mean of -27.2 ¡C) and March colder by 1.6¡C (mean of -23.1¡C). These differences are evident on Figures 3. But the main difference in thermal behavior lies in the fact that 1990-91 had long cold spells of 2-4 weeks duration separated by warmer periods of shorter duration. Openings took place mainly in the lows of the cold spells, generally after many hours (roughly 4 days) of soil cooling, but the cracks then closed and opened again with secondary temperature oscillations in the range of 12 to 48 hours. Michel Allard, Jennifer N. Kasper 9 Discussion Each winter, a majority of cables (7 in 1989-91 ; 8 in 1990-91) spread across the cracks were visibly strained although all did not break. Only cables 8 and 10 were not strained in either year. At the start, these two cables were laid across apparent lines of fractures of dubious origin, perhaps resulting from slope tension in the vegetation on the inner side of a furrow (#10) or at a former site of cracking that has become inactive (#8). The small spatial and temporal scale of the observations do not attain the statistical significance of MackayÕs (1974, 1992) frequency data spread over tens of frost cracks, at three sites over many years. However the analysis of thermal conditions when tension occurred on the wires provide numerical values that substantiate the previous theoretical work (Lachenbruch, 1962, 1966). In both years cracking began in late December-early January when air temperatures dropped from about -20¡C to below -32¡C, after four months of gradual cooling. First and subsequent crack openings took place at similar soil and atmospheric temperatures in both winters. For instance, average temperatures at opening times at the wedge top were -20.0¡C in 1989-90 and -19.7¡C in 1990-91. The difference between the two winters is in the response time to oscillations and cooling rate before crack openings. While in 1989-90 cracks opened after temperature drops of about 34 hours duration at a mean rate of -14¡C/day, they did so the following year after 102 hours at a rate of -3.7¡C/day. This observation applies also to the soil; in response to temperature changes at 8 cm, cracks opened after drops lasting about 60 hours at a rate of -1.7¡C/day the first winter and lasting over twice as long (134 hours) the following year at about half the rate (-0.9¡C/day). In fact the ratio of the difference is approximately 3 in the air between the two winters, (i.e. roughly three times longer at 1/3 of the cooling rate ; at 8 cm, it is about 2. Mackay (1993) calculated that ice-wedge cracks on Garry Island usually open after four days of decreasing air temperatures (at a rate of -1.8¡C/day), a situation that resembles our data for winter 1990-91 but differs greatly from 1989-90. His study used daily mean air temperatures from Tuktoyaktuk, 80 km away, which brought some generalization into the data . The present study shows that the thermal regime at the site governs expansion and contraction. Comparison between the two winters suggests that synoptic weather variations (duration and interval of passages of colder air masses) regulated the rhythm of cracking over the two years. During the observation period, the damping of temperature variations in the soil by snowcover did not have a serious impact on frost cracking (Mackay, 1978, 1993). The surface and shallow temperatures were warmer in the polygonal trough than in the polygon 10 centre in early winter on both years; however they cooled down to the same values over the second half of January. Despite the absence of snowcover observations, this behavior can easily be explained by the filling of the trough and of the depressed polygon center by 30-40 cm of fresh snow in early winter, which was not enough to prevent the soil from cooling to cracking temperatures. Thereafter, wind prevented further accumulation on the site (average wind velocity of 5.3 m/s, recorded maximum of 24 m/s ; data not shown). Almost all of the closing events recorded are probably true events and not instrumental artifacts. For instance, it is worth noting that many closing events took place when the temperature in the soil was colder than at the time of earlier crack openings. Therefore, expansion and contraction of the metallic wires inside their sheeting is not the cause of circuit closing. The authenticity of the events is also supported by the straightforward relationship with temperature variations in all cases in the first winter and in the majority of cases in the second one. In both years, and despite different temperature regimes governing cracking, closing events occurred either after a warming of 11-13¡C over a period of 27-30 hours or during oscillations within the course of long very cold spells. They may be related to ground movements generated elsewhere in the frozen ground stress field. In general, the occurrence of closings in response to temperature changes of almost uniform amplitude and duration suggests that, once cracked, the permafrost and frozen active layer expand and contract regularly in response to temperature variations. These expansion and contraction movements logically should take place in stress fields bounded by already open cracks, i.e., within tundra polygons. Reopening or enlarging of existing cracks with falling temperatures may possibly be related with the lateral expansion of cracks along wedges. For example, Mackay (1984) reports from an experiment at Garry Island that the same wedge cracked on three different dates over ten days along a propagation distance of 5 m. Each new cracking may then very well provoke the enlargement of the already open crack upstream along the direction of propagation. In May 1990, crack #4, which last opened on January 20, closed as the temperature increased (-9¡C at 8 cm to -12¡C at 258 cm) and as the thermal gradient in the ground was reversing. This late closing is likely another demonstration of the effect of frozen ground expansion. It has been demonstrated elsewhere that wedge cracks can close by as much of 80 % between their maximum width in winter and early summer (Mackay, 1975). However, in many excavations in the area around the site, we found several cracks in wedges below the permafrost table in July and August that were still wide The 7th International Permafrost Conference enough to introduce a thin (Swiss army) knife blade. Others were completely sealed by new ice veins and some had corks of ice and soil that seamed their upper sections. Conclusion The temperatures conditions that were measured when thermal cracks opened are in excellent agreement with existing theoretical work. Air temperature oscillations duringthe two years of observation, certainly related to synoptic weather variations, regulated the rate of cooling and the time response of ice-wedge mechanical reactions. However, crack opening always took place at the same well-defined ground temperature. These field measurements allow one to estimate winter conditions of past climate in regions possessing icewedge pseudomorphs and in permafrost regions where ice-wedges have become dormant (i.e. inactive) due to a change in climate. One must keep in mind, however, that cracking temperatures can occur in locally colder environments within the discontinuous permafrost zone (e.g., Burn, 1990; Hamilton et al., 1983; Payette et al., 1986) Acknowledgments The authors express their thanks to Dr. Richard Fortier who designed and built the automatic meteorological station and the crack detection circuit and for reading a first draft. Maintenance and data handling were subsequently carried on by Dr. Janusz Frydecki. This study received the financial and logistical support of Natural Sciences and Engineering Research Council of Canada (Allard), Fonds pour la Formation de Chercheurs et lÕAide ˆ la Recherche du Minist•re de lÕEnseignement SupŽrieur du QuŽbec (Allard), the Geological Survey of Canada, the Polar Continental Shelf Program of Canada and the Northern Studies Training Program of the Department of Indian Affairs and Northern Development (Kasper). Reviewing by Dr. C. R. Burn and another, anonymous, referee greatly helped to improve the manuscript. Special thanks are expressed to the community of Salluit which hosted us and provided true friendship during field work. References Allard, M., Tremblay, C., Pilon, J.A. and Frydecki, J. (1993). Quaternary geology and geocryology in Nunavik, Canada. In Proceedings, Sixth International Conference on Permafrost, Beijing, China, Vol. 1, pp. 5-10. Lachenbruch, A.H. (1966). Contraction theory of ice-wedge polygons : A qualitative discussion. In Proceedings, First International Permafrost Conference, National Academy of Sciences-National Research Council, Washington, D.C., publication no 1287, pp. 63-70. Allard, M., Wang, B. and Pilon, J.A. (1995). Recent cooling along the southern shore of Hudson Strait, QuŽbec, Canada, documented from permafrost temperature measurements. Arctic and Alpine Research, 27, 157-166. Leffingwell, E. de K. (1919). The Canning River Region, Northern Alaska. United States Geological Survey Professional Paper, 109, 251 p. Burn, C.R. (1990). Implications for palaeoenvironmental reconstruction of recent ice-wedge development at Mayo, Yukon Territory. Permafrost and Periglacial Processes, 1, 3-14. Mackay, J.R. (1974). Ice-wedge cracks, Garry Island, Northwest Territories. Canadian Journal of Earth Sciences, 11, 1366-1383. Gray, J.T. (1983). Extraction and compilation of available temperature and snowfall data in the Ungava Peninsula as input to geothermal modelling of Quaternary paleoclimates. Earth Physics Branch, Energy, Mines and Resources Canada, Open File, Report MAS 2050-2-1385, 33 p. Mackay, J.R. (1975). The closing of ice-wedge cracks in permafrost, Garry Island, Northwest Territories. Canadian Journal of Earth Sciences, 12, 1668-1674. Grechishchev, S.Y.E. (1973). Basic laws of thermorheology and temperature cracking of frozen ground. In USSR Contribution, Permafrost, Second International Conference, National Academy of Sciences, Washington, D.C., pp. 228234. Mackay, J.R. (1978). The use of snow fences to reduce icewedge cracking, Garry Island, Northwest Territories. In Current Research, part A, Geological Survey of Canada, Paper 78-1A, pp. 523-524. Mackay, J.R. (1984). The direction of ice-wedge cracking in permafrost : upward or downward ? Canadian Journal of Earth Sciences, 21, 516-524. Hamilton, T.D., Ager, T.A. and Robinson, S.W. (1983). Late Holocene ice wedges near Fairbanks, Alaska, U.S.A. : environmental setting and history of growth. Arctic and Alpine Research, 15, 157-168. Mackay, J.R. (1992). The frequency of ice-wedge cracking (1967-1987) at Garry Island, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 29, 236-248. Kasper, J.N. (1995). Geomorphic, geophysical and Quaternary studies of ice and soil wedge features in the Foucault River Valley, Northern QuŽbec. UniversitŽ Laval, Department of Geography, Ph. D. thesis, 277 p. Mackay, J.R. (1993). Air temperature, snow cover, creep of frozen ground, and the time of ice-wedge cracking, western Arctic coast. Canadian Journal of Earth Sciences, 30, 17201729. Lachenbruch, A.H. (1962). Mechanics of thermal contraction cracks and ice-wedge polygons in permafrost. Geological Society of America, Special Paper, 70, 69 p. Osterkamp, T.E. (1974). Temperature measurements in permafrost. Alaska Department of Transportation and Public Facilities, Report No. TFHWA-AK-RD-85-11. Michel Allard, Jennifer N. Kasper 11 Payette, S., Gauthier, L. and Grenier, I. (1986). Dating icewedge growth in subarctic peatlands following deforestation. Nature, 322, 724-727. PŽwŽ, T.L. (1966). Ice-wedges in Alaska : Classification, distribution and climatic significance. In Proceedings, First International Permafrost Conference, National Academy of Sciences-National Research Council, Washington, D.C., publication no 1287, pp. 76-81. 12 The 7th International Permafrost Conference